Presentations

Abstract

Evaluating wind farm control strategies requires a careful selection of simulation methods. The decision typically lies between computationally expensive models and more efficient alternatives. The former covers high-fidelity aeroelastic models combined with computational fluid dynamics (CFD), enabling the evaluation of mechanical loads under dynamic conditions. In contrast, the latter includes faster steady-state wake models, such as FOXES [1] or FLORIS [2], which utilize actuator disks to evaluate the power performance of wind farms. To achieve the required computational speed, steady-state wake models filter out high frequencies from the wind, which is a valid approach for actuator disk models. However, to utilize high-fidelity aeroelastic tools alongside fast wake models—taking advantage of their speed while still being able to conduct load analyses—the lost high-frequency information must be synthetically restored. This work presents a method for generating turbulent wind fields from steady-state wake models by expanding the Continuous-Time Random Walk (CTRW) model developed by David Kleinhans [3]. The CTRW model is an alternative to spectral turbulence models like the Mann or Kaimal model. One major advantage of the CTRW model is its ability to generate wind fields indefinitely alongside a time domain simulation by solving its stochastic differential equations. For this reason, the CTRW model has been implemented in the aeroelastic tool MoWiT [4], where it serves as an input for real-time simulations. This implementation allows for continuous feeding of wind speed and turbulence at hub height into the model, creating full turbulent wind fields in front of the wind turbine. Building upon this, the proposed method expands the CTRW model by incorporating additional input variables retrieved from the quasi-steady-state wake model FOXES. This includes local wind speed and turbulence intensity from the sheared ambient wind and the wake at each grid point of the wind field. This reduces uncertainties regarding the spatial distribution of wind speed and turbulence intensity along the grid, which was only approximated by fitted parameters in the original model. The plausibility of the generated turbulence is evaluated by analyzing common statistical characteristics, with particular emphasis on the effects of incorporating wake deficits as inputs to the CTRW model.  The proposed method offers a straightforward solution for coupling existing aeroelastic tools, which are not yet compatible with a wind farm environment, with open-source steady-state wake models. This approach provides an accessible entry point into wind farm simulation, enabling the testing of wind farm controllers designed to optimize loads. [1] Schmidt, J. et al., (2023). FOXES: Farm Optimization and eXtended yield Evaluation Software.  Journal of Open Source Software, 8(86), 5464, https://doi.org/10.21105/joss.05464  [2] FLORIS: FLOw Redirection and Induction in Steady State, https://www.nrel.gov/wind/floris.html [3] Kleinhans, D. (2008). Stochastische Modellierung komplexer Systeme: von den theoretischen Grundlagen zur Simulation atmosphärischer Windfelder [4] MoWiT: Modelica Library for Wind Turbines, www.mowit.info